[Click] grid DSDV + nsclick

Douglas S. J. De Couto decouto at csail.mit.edu
Tue Feb 10 10:41:05 EST 2004


> I'm going to take a look at the code one of the next days, but I want 
> to
> check if everything is working. I've been looking at
> click-dir/conf/simple-dsdv-userlevel.click, and removed some 
> references to
> non-nsclick elements, and tried to replace them with the appropriate 
> ones.
>

> The comments mention build-grid-config.sh, which will probably fill in
> some of the variables for DSDVRouteTable, etc. I can't find the file
> anywhere (I've checked click-dir/tools/). Could somebody tell me where 
> I
> can find this file, or just give me some default values for
> DSDVRouteTable?

You should use make-dsdv-config.pl, which should produce an actual 
working click config (with all the appropriate element parameters) 
which you can then examine (or look at the script source as well).  
build-grid-config.sh has gone to the big shell script directory in the 
sky, and i am keeping those other DSDV configs around only as 
historical reference.


> Secondly, the element KernelTun is used in the click-file. Apparently 
> this
> is not available for nsclick. Can I just replace them with
> FromSimDevice(tap0) and ToSimDevice(tap0) or is that wrong?

KernelTun is what hooks up a userlevel click router to the local 
machine's OS so that userlevel processes can send data via click.  You 
should use whatever nsclick element is appropriate for that....

> And last, I see a lot of references in the comments about Grid 
> packets. To
> be honest I don't really understand 100% what they are used for. Are 
> they
> some kind of protocol packets? Can I still have my own packets routed 
> by
> the DSDV implementation?

`Grid' is just the single name I use to refer to everything having to 
do with my protocol implementations.  In the DSDV implementation, all 
protocol and data packets are `Grid' packets, and have an ethertype of 
0x7fff or 0x7ffe.  IP Data packets are encapsulated in a Grid header 
which adds some routing information.  They are unencapsulated at the 
receiver.


> I know I'm asking a lot, but any help is appreciated.

No problem!

d
--
Douglas S. J. De Couto    <decouto at csail.mit.edu>



More information about the click mailing list